First Pass

10 stories from 6 sources

AI’s next contest is infrastructure, control, and trusted use

Day’s Recap

Supporting Articles

3:05 AMTechCrunch

Meta signs first AI data center deal in India with Reliance

Summary

Meta signed a deal with Reliance to secure capacity in a 168-megawatt data center in India to support its global AI computing needs, with room to expand. The agreement gives Meta a new large-scale footprint outside its traditional North American and European infrastructure base.

Why it matters

AI advantage is increasingly constrained by power and data center access, and this deal secures both while repositioning India as a meaningful node in global AI compute.

Link to Story
9:35 PMPYMNTS

Visa Launches AI and Stablecoin Tools to Power Agentic Commerce

Summary

Visa unveiled a set of new AI, stablecoin, and token initiatives it says will help clients enable faster, more automated, more intelligent commerce. The company positioned the release as infrastructure for agentic commerce, where software agents can initiate and complete transactions.

Why it matters

If Visa succeeds, it can keep its network central as commerce shifts from human checkout to software-driven purchasing and stablecoin-enabled settlement.

Link to Story
1:20 PMThe Verge

Google won’t just admit it’s feeding YouTube creators to its music AI

Summary

Independent musicians allege Google used songs uploaded to YouTube to train its Lyria 3 music model without permission. Google’s court posture suggests it views YouTube uploads as available for training, while avoiding a direct public admission.

Why it matters

The case could set the practical rule for whether uploading art to a major platform also licenses it to train AI.

Link to Story

Other Developments

A curated list of other prominent stories from this day.

8:01 PMFinextra

Capsa AI raises $18m to solve private capital's data problem

Summary

Capsa AI has raised $18 million in Series A funding to build an AI operating system aimed at private capital firms. The product targets fragmented, hard-to-use data across the private markets investment lifecycle.

Why it matters

Private markets run on messy, proprietary information, so the company that standardizes it can win durable distribution and pricing power across deal and portfolio workflows.

Link to Story
6:52 PMThe Verge

Apple’s new Siri AI knows when to shut up

Summary

Apple has begun rolling out a new Siri AI experience that prioritizes brief, direct answers over chatty conversation. Early use suggests it feels more reliable in everyday interactions because it avoids unnecessary verbosity.

Why it matters

If Siri can deliver accurate answers with minimal output, Apple can make voice AI feel like a utility again and reduce user dependence on standalone chat apps.

Link to Story
1:07 PMTechCrunch

‘AI-pilled’ firms spend $7,500 per employee each month on AI

Summary

The most AI-committed companies are spending about $7,500 per employee per month on AI tools and infrastructure, according to Ramp’s AI Index. That level of spend is rising, but it still sits below the rough cost of adding another engineer.

Why it matters

AI is moving into core opex, forcing companies to justify cost per employee the same way they justify headcount.

Link to Story
9:07 AMTechCrunch

Decart’s new world model can simulate hours of photorealistic driving — with some caveats

Summary

Decart released Oasis 3, a real-time world model that generates photorealistic driving simulations for autonomous vehicle development and testing. The system is now available via an API so developers and AV teams can integrate the simulator into their workflows.

Why it matters

Cheaper, faster photoreal simulation can compress AV development cycles, but only if teams can trust the simulator enough to use it for safety-relevant decisions.

Link to Story
6:52 AMScienceDaily

A classic brain test exposed AI's biggest weakness

Summary

Top AI models struggled on a standard psychological attention test as the task length and complexity increased. Accuracy dropped from above 90% on short sequences to near failure on longer versions, suggesting brittle attention control under sustained load.

Why it matters

It flags a reliability ceiling for current models in long, complex workflows where small attention slips can compound into total failure.

Link to Story
5:29 AMBloomberg Markets

Anthropic Releases New Mythos-Like Model Ahead of IPO | The Opening Trade 6/10/2026

Summary

Anthropic released its new model, Fable 5, with built-in blocks on answering questions related to cybersecurity and biology. The launch comes as the company pushes toward an IPO amid intensifying competition with other major AI developers.

Why it matters

The product sets a clearer market line between maximum capability and safety-constrained access, which will shape enterprise procurement and IPO narratives.

Link to Story
5:00 AMThe Verge

Waymo built a virtual driver to study how humans react to surprises on the road

Summary

Waymo built a virtual, hyperattentive driver inside simulated 3D worlds to test how people respond to unexpected road events and to benchmark its autonomous system against human-like behavior. The work extends Waymo's use of simulation to probe rare edge cases and chaotic scenarios that are hard to capture reliably in real-world driving.

Why it matters

Better human-behavior simulation can accelerate autonomy rollout, but it also raises the stakes on whether Waymo's assumptions about people are accurate enough to support safety and policy decisions.

Link to Story

Make it yours

Build Your First Pass.

Pick your topics, set your cadence, and receive your personalized First Pass in your inbox. It’s that simple!